News PowerColor RX 5700 ITX

According to this article on OC3D.net PowerColor is coming out with a RX 5700 itx card, very good news. Limited to the Japanese market but hopefully someone can get their hands on one.




Edit: There is a VBIOS the brings the UEFI GOP to the 5700 ITX. You can find it here: https://www.techpowerup.com/vgabios/223333/223333
Credit @Soul_Est
 
Last edited:

Zgembo

Trash Compacter
Apr 14, 2017
53
108
I'm very software-technical (programmer for a living) but not familiar with BIOS flashing; does the fact that, for example, the Sapphire Pulse has 2 power connectors versus 1 in the PowerColor ITX not influence the VBIOS you're flashing?
It should not. What it may affect is supported connectors, for example; that Sapphire card has 3x DisplayPort & 1x HDMI connectors, where our ITX card only has 2x DisplayPort & 1x HDMI. Anyway, both of my DisplayPort ports do work fine, I guess HDMI works as well, but I can't really check as I have nothing to plug in there.
 

paradigmpost

Chassis Packer
Feb 26, 2019
16
13
Got it and installed; very happy with it with the exception that I'm noticing weird system lock ups when coming back from sleep (consistently) and when viewing GPU accelerated video (has happened twice now); no issues in gaming. Not sure if this is some other configuration issue (doubtful as it just started, but possible). Going to give a reference card ROM a shot as @Zgembo did and see if that improves the situation.
 

paradigmpost

Chassis Packer
Feb 26, 2019
16
13
Since flashing the VBIOS, I have to completely shut off GPU acceleration in Google Chrome (or Chrome-based browsers) or it opens to a black screen. I've also tweaked the fan curve to have it run a bit cooler as I noticed it at pretty high temperatures at what should've been relatively close to idle (again, since flashing the VBIOS). Other than that, everything works great and performance is good. I may flash back to the old VBIOS to do some more comparisons. If I could figure this stuff out, I'd be ecstatic about this card.

Wondering if you've noticed anything finicky with your card @Zgembo?
 

Zgembo

Trash Compacter
Apr 14, 2017
53
108
Wondering if you've noticed anything finicky with your card @Zgembo?
Now that you ask, I actually do notice some funky things. With original VBIOS I had 0 issues (other than missing GOP that is), but with Sapphire one I do have occasional system freeze, where kernel would spawn something like (only visible via ssh session, as display output dies completely):
[drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=n, emitted seq=n+2
(where n is some random large number) which totally relates to infamous https://gitlab.freedesktop.org/drm/amd/issues/892, which is super similar to infamous Windows "black screen" bug. I'll be reverting to previous VBIOS just to confirm it ain't related to other stuff...
 

dss

Efficiency Noob
Feb 8, 2020
7
6
For folks with this card, how are the temps/fan noise? Does the fan turn off at idle? Thanks.
 

Zgembo

Trash Compacter
Apr 14, 2017
53
108
Since flashing the VBIOS, I have to completely shut off GPU acceleration in Google Chrome (or Chrome-based browsers) or it opens to a black screen. I've also tweaked the fan curve to have it run a bit cooler as I noticed it at pretty high temperatures at what should've been relatively close to idle (again, since flashing the VBIOS). Other than that, everything works great and performance is good. I may flash back to the old VBIOS to do some more comparisons. If I could figure this stuff out, I'd be ecstatic about this card.

Wondering if you've noticed anything finicky with your card @Zgembo?
I've reverted to original VBIOS, and the card has been rock solid, not a single crash or freeze. Keeping it this way.
 

Soul_Est

SFF Guru
SFFn Staff
Feb 12, 2016
1,531
1,926
Myself and one other person have inquired about a possible BIOS update for this card to correct the issue. With the UEFI GOP in place, this card would be pretty much perfect.

This card seems to be for the Japanese market specifically. If you look on Amazon Japan, there are a number of cards that have the typical PowerColor design and are branded as Genjin Oriented ().

We'll see what the response is once either of us get a reply.
 
Last edited:

Soul_Est

SFF Guru
SFFn Staff
Feb 12, 2016
1,531
1,926
I reached out to someone who has a PowerColor RX 5700 (non-ITX) to see if I can get a copy of the BIOS from that card for comparison.
 

Zgembo

Trash Compacter
Apr 14, 2017
53
108
I reached out to someone who has a PowerColor RX 5700 (non-ITX) to see if I can get a copy of the BIOS from that card for comparison.
Careful there, couple weeks ago I flashed my PowerColor RX5700 ITX to https://www.techpowerup.com/vgabios/213952/powercolor-rx5700-8192-190616 VBIOS, resulting in bricked card. It took VERY significant effort to recover the card, including SPI chip 1+8 pin "trick" (actually shorting SPI flash pin 1 aka CS to ground was a must otherwise the system wouldn't even post) and then forcing system to & back from sleep (or was it hybernation?) in order for flasher to work, otherwise the eeprom wasn't detected at all (asic gets detected fine by amdvbflash, but flash chip does not, amdvbflash sais: "Unknown SPI", and the force flash results in ERROR: 0FL01). And all of that WITHOUT second VGA, via remote screen session :|. It looks like PowerColor reference RX5700 uses different GDDR6 memory.
 
Last edited:

Soul_Est

SFF Guru
SFFn Staff
Feb 12, 2016
1,531
1,926
Careful there, couple weeks ago I flashed my PowerColor RX5700 ITX to https://www.techpowerup.com/vgabios/213952/powercolor-rx5700-8192-190616 VBIOS, resulting in bricked card. It took VERY significant effort to recover the card, including SPI chip 1+8 pin "trick" (actually shorting SPI flash pin 1 aka CS to ground was a must otherwise the system wouldn't even post) and then forcing system to & back from sleep (or was it hybernation?) in order for flasher to work, otherwise the eeprom wasn't detected at all (asic gets detected fine by amdvbflash, but flash chip does not, amdvbflash sais: "Unknown SPI", and the force flash results in ERROR: 0FL01). And all of that WITHOUT second VGA, via remote screen session :|. It looks like PowerColor reference RX5700 uses different GDDR6 memory.
Thank you for the fair warning. I was looking at a PowerColor designed and engineered model (https://www.powercolor.com/product?id=1565845929) and not a reference model for better or the best compatibility.
 

Zgembo

Trash Compacter
Apr 14, 2017
53
108
Well, it is up to you, if my card was more accessible (it is quite stuck under all those heatpipes & DB4), I might even try to flash it again now that I have the proven recovery procedure, but it is just too much the hassle to take that risk. If you are willing to try, make sure you have backup VBIOS somewhere safe & PC capable of running two graphics cards or APU in hand, as well as remote desktop enabled. Heck, I'd even keep an external eeprom flasher nearby, just in case.
 

Mackan

Airflow Optimizer
Jun 2, 2016
302
160
I contacted Powercolor and they said their engineer is working on a new VBIOS. So they have acknowledged the problem with no UEFI booting for this card. They would get back to me when the new VBIOS was done. But it's been several weeks now, so something is taking time.

As for the new 5600 XT, no way to know if this one suffers from the same problem or not, unless a reviewer discovers it (doubtful), or someone buys it and confirms.

Very good that they keep up the habit to release ITX cards though.
 

Basriy

Chassis Packer
Feb 18, 2020
14
32
I
I've received the card, but I won't be home to give it a whirl until Friday night... the suspense is killing me.



Not excited to see what that means for the dual-boot hackintosh I planned to put this in; hopefully it's okay.



I've been having trouble finding anything about this attached to PowerColor other than news articles, and they don't have a Japanese site that I can find. Hoping they'll at least humor this kind of request via email.
I am also planning to use it in Hackintosh. Have ordered just recently. Waiting.